TOEFL Speaking practice and strategy

How the Speaking section is built

Speaking is the shortest section, at about 8 minutes of base testing time, and it contains only two task types. Like Writing, it is linear, so everyone gets the same tasks. Unlike every other section, it records you, and nothing you record can be changed.

The defining constraint is that there is no preparation time. Not a short amount: none. In both tasks you respond immediately, which means the section cannot be approached the way many test takers were taught to approach older speaking formats, where planning time was part of the design.

You also cannot re-record. If you stumble mid-answer you can correct yourself out loud, exactly as you would in conversation, and that costs you very little. Stopping to start again is not available.

Listen and Repeat

You are placed in a realistic scenario such as helping out somewhere on campus, showing people around, or being trained for a task, and you hear seven sentences in that context. You repeat each one, once, immediately after hearing it. A visual accompanies the scenario and updates as you progress, which keeps the task grounded in a situation rather than presenting it as a list of drills.

The sentences get harder in a deliberate progression. The first two are short, around 9 to 11 syllables, usually a single clause, with 8 seconds to respond. Three medium sentences follow at roughly 14 to 16 syllables, often carrying a prepositional phrase or a dependent clause, with 10 seconds. The last two are the demanding ones, at about 19 to 23 syllables with complex syntax, serial lists and multiple modifiers, and you get 12 seconds each.

Scoring is holistic, on a 0 to 5 scale, against two criteria: intelligibility, meaning how clearly you pronounce what you say, and accuracy, meaning how closely your repetition matches what you heard. Crucially, partial credit exists. Reproducing part of a sentence accurately is worth something, so an attempt that trails off still scores above silence.

• Wait for the beep. Speaking before it can cost you the start of your response.
• Finish the sentence even after a mistake, because partial accuracy still earns credit.
• Hold the meaning, not just the sounds. Understanding a sentence makes it far easier to reproduce.
• Speak at your normal pace. Rushing to fit the window damages intelligibility, which is half the score.

Take an Interview

You answer four questions in a simulated interview, with 45 seconds for each and no preparation time. The questions range across the personal, the abstract and the speculative: what you would do, what you think, why something might be the case. There is no fixed subject to revise.

Forty-five seconds is longer than it feels when you begin and shorter than it feels when you are halfway through. The practical shape that fits it is simple: answer the question in your first sentence, then spend the remaining time supporting that answer with a reason and something concrete. Test takers who build up to their answer frequently run out of time before arriving at it.

If you finish early, keep going and add a detail, an example, or a qualification. Trailing into silence with fifteen seconds left leaves marks unclaimed. Repeating yourself does not help, but genuinely developing the point does.

• Answer in the first sentence. Do not build up to it.
• Then support: one reason, one concrete example.
• Use the full window. If you finish early, elaborate rather than stop.
• Correct yourself out loud if you slip, because that is what real speakers do and it costs little.

How Speaking is scored

Responses are scored holistically: one score for the response as a whole rather than a tally of errors. The criteria vary a little by task, but the recurring ones are delivery, language use, and elaboration: how clear the speech is, how effectively grammar and vocabulary carry the ideas, and how well the ideas are supported. Relevance sits underneath all of it: an eloquent answer to a different question scores badly.

Because the speech is spontaneous, occasional errors are expected and do not prevent a high score. This is worth internalising, because the fear of making mistakes produces the two habits that actually lower scores: long pauses while you search for a better word, and sentences abandoned halfway through.

Accent is not scored. A natural accent is expected and affects nothing unless it makes you hard to understand. Intelligibility is the standard, not resemblance to any particular variety of English.

• Delivery: clear, reasonably paced, natural intonation.
• Language use: grammar and vocabulary that carry the meaning, not that show off.
• Elaboration: ideas supported rather than asserted.
• Relevance: answering the question that was actually asked.

Practising a section that records you

Speaking is the section learners most often practise wrongly, because it is the least comfortable to hear back. Practice without recording is close to worthless here: you cannot hear your own filler, your own pauses, or the sentence you left unfinished while you were producing it.

So record, then listen with a fixed set of questions. Did I actually answer what was asked? Was I clear throughout? Did I use the whole window? Where did I hesitate, and was it a word I did not know or a thought I had not formed? The distinction between those last two matters, because one is a vocabulary problem and the other is a planning problem, and they need different work.

Practise under the real constraint. Answering with ten seconds of thinking time first rehearses a task that does not exist. Hear the question, wait for the beep, and start.

• Always record. Reviewing the audio is where the improvement happens.
• Practise with no preparation time, because there is none.
• Track fillers and long pauses across sessions rather than within one.
• Check your finish: did you use the full window, or stop early?

Microphone and setup

Delivery is scored from a recording, so the recording has to be usable. Check your microphone and volume before you begin, speak at a normal conversational level, and keep the microphone near enough to pick you up clearly without being right against your mouth, which distorts more than it captures.

Practise on the equipment and in the room you expect to use. A headset that clips your first syllable, or a room with an echo, will cost you marks in a way that has nothing to do with your English, and it is much better to discover that during practice.

• Test the microphone and volume before starting.
• Speak at a natural volume, neither whispering nor pressed against the mic.
• Use a quiet space and silence notifications.
• Rehearse on your real setup so problems surface early.

How our Speaking practice is made

Our scenarios, sentence sets and interview questions are written and recorded by us, built to the current task structure. Our Listen and Repeat practice gives you 8 seconds to record each response and matches a server transcript of your recording against the target sentence; interview responses can optionally be sent for evaluation that comments on delivery, language use and development.

Neither is an official score. They are study tools for spotting patterns across attempts, such as a recurring pause or a structure you always avoid, and should be read that way. Format facts on this page come from ETS and are linked below.